The effect of confounding data features on a deep learning algorithm to predict complete coronary occlusion in a retrospective observational setting.

Aims: Deep learning (DL) has emerged in recent years as an effective technique in automated ECG analysis. Methods and results: A retrospective, observational study was designed to assess the feasibility of detecting induced coronary artery occlusion in human subjects earlier than experienced cardiologists using a DL algorithm. A deep convolutional neural network was trained using data from the STAFF III database. The task was to classify ECG samples as showing acute coronary artery occlusion, or no occlusion. Occluded samples were recorded after 60 s of balloon occlusion of a single coronary artery. For the first iteration of the experiment, non-occluded samples were taken from ECGs recorded in a restroom prior to entering theatres. For the second iteration of the experiment, non-occluded samples were taken in the theatre prior to balloon inflation. Results were obtained using a cross-validation approach. In the first iteration of the experiment, the DL model achieved an F1 score of 0.814, which was higher than any of three reviewing cardiologists or STEMI criteria. In the second iteration of the experiment, the DL model achieved an F1 score of 0.533, which is akin to the performance of a random chance classifier. Conclusion: The dataset was too small for the second model to achieve meaningful performance, despite the use of transfer learning. However, 'data leakage' during the first iteration of the experiment led to falsely high results. This study highlights the risk of DL models leveraging data leaks to produce spurious results.

The "bladder sign"--an important early marker of retroperitoneal hemorrhage.

Retroperitoneal hemorrhage remains one of the major complications of cardiac and peripheral vascular catheterization. Its high associated morbidity and mortality require vigilance and early intervention. We report six cases of retroperitoneal hemorrhage featuring a "bladder sign." The compression of the bladder described in this series can be visualized on the incidental cystogram that results from contrast given during catheterization. Its significance as a highly specific marker of retroperitoneal hemorrhage should be appreciated.

Clinical characteristics of tako-tsubo cardiomyopathy.

Tako-tsubo cardiomyopathy (TTC) is increasingly diagnosed in the United States, especially in the Caucasian population. To evaluate the clinical features and outcome of patients with TTC, we evaluated 34 patients (32 women and 2 men) 22 to 88 years of age (mean 66 +/- 14) who fulfilled the following criteria: (1) akinesia or dyskinesia of the apical and/or midventricular segments of the left ventricle with regional wall motion abnormalities that extended beyond the distribution of a single epicardial vessel and (2) absence of obstructive coronary artery disease. Twenty-five patients (74%) presented with chest pain, 20 patients (59%) presented with dyspnea, and 8 patients (24%) presented with cardiogenic shock. Twenty-two patients (65%) had ST-segment elevation and 14 patients (41%) had T-wave inversion on presentation. Twenty-five patients (74%) reported a preceding stressful event. Cardiac biomarkers were often mildly increased, with a mean troponin I (peak) of 13.9 +/- 24. Mean +/- SD left ventricular ejection fractions were 28 +/- 10% at time of presentation and 51 +/- 14 at time of follow-up (p <0.0001). Two patients (6%) died during the hospital stay. Average duration of hospital stay was 6.6 +/- 6.2 days. In conclusion, TTC is common in postmenopausal women with preceding physical or emotional stress. It predominantly involves the apical portion of the left ventricle and patients with this condition have a favorable outcome with appropriate medical management. The precise cause remains unclear.

Prevalence of metabolic syndrome in patients < or =45 years of age with acute myocardial infarction having percutaneous coronary intervention.

The prevalence of metabolic syndrome (MS) was determined in patients aged < or =45 years who presented with acute myocardial infarction and underwent primary percutaneous coronary intervention. Two hundred twenty-three consecutive patients aged 18 to 45 years who underwent cardiac catheterization for acute myocardial infarction from June 2001 to December 2004 were reviewed. MS was diagnosed by National Cholesterol Education Program Adult Treatment Panel III guidelines (modified by substituting body mass index > or =28.8 kg/m2 for waist circumference). One hundred sixty-one patients met all 5 criteria for MS available for evaluation. Seventy-six of these patients (47%) met > or =3 of the 5 criteria for MS. Sixteen patients with MS (21%) and 5 patients without MS (6%) had diabetes mellitus. The prevalence of each criterion was significantly higher (p <0.05) in the MS group. Average Framingham risk scores were 7.0 and 4.5 for patients with and without MS, respectively. The prevalence of smoking, male gender, and family history of premature coronary artery disease were the same for the 2 groups. In conclusion, MS was highly prevalent in this population of young patients with acute myocardial infarction.

Stroke complicating percutaneous coronary intervention in patients with acute myocardial infarction.

BACKGROUND: Stroke associated with percutaneous coronary intervention (PCI) is a tragic complication. Despite advances in the practice of PCI, the incidence of stroke complicating PCI has not changed over the decades. The objective of the present study was to evaluate incidence and correlates of stroke occurring in patients with myocardial infarction (MI) undergoing PCI. METHODS AND RESULTS: Stroke was defined as the presence of any new focal neurological deficit lasting > or =24 h that occurred anytime during or after PCI until discharge. In 2,281 consecutive patients with PCIs for non-ST-elevation MI, or ST-elevation MI (STEMI), 20 strokes were identified (0.88%). Strokes were ischemic in 95%. On multivariate analyses, ejection fraction < or =30% (odds ratio =4.3, p=0.003) was the only independent predictor for stroke. In patients who developed stroke within 24 h of PCI, PCI of vein grafts was more frequent, and use of glycoprotein IIb/IIIa inhibitor was less frequent. Those patients tended to present late in the course of MI. Stroke found more than 24 h after PCI was related to diabetes, higher serum creatinine, lower ejection fraction, anterior wall STEMI and emergency use of intra-aortic balloon pumps. CONCLUSIONS: Low ejection fraction was the only independent predictor for stroke, but risk factors for periprocedural stroke are different from those of stroke occurring more than 24 h after PCI. Upstream use of glycoprotein IIb/IIIa inhibitor might decrease the risk of periprocedural stroke.

Retroperitoneal hemorrhage from inferior epigastric artery: value of femoral angiography for detection and management.

Laceration of the inferior epigastric artery during cardiac catheterization is under-reported in the literature, but it is a serious complication leading to retroperitoneal hemorrhage and even death. We report on two cases of retroperitoneal bleeding from unintentional puncture of the inferior epigastric artery during cardiac catheterization. Femoral angiography is a valuable tool to promptly recognize this complication.

Myocardial infarction as a presentation of clinical in-stent restenosis.

BACKGROUND: In-stent restenosis is considered to be a gradual and progressive condition and there is scant data on myocardial infarction (MI) as a clinical presentation. METHODS AND RESULTS: Of 2,462 consecutive patients who underwent percutaneous coronary intervention between June 2001 and December 2002, clinical in-stent restenosis occurred in 212 (8.6%), who were classified into 3 groups: ST elevation MI (STEMI), non-ST elevation MI (NSTEMI) and non-MI. Of the 212 patients presenting with clinical in-stent restenosis, 22 (10.4%) had MI (creatine kinase (CK)>or=2xbaseline with elevated CKMB). The remaining 190 (89.6%) patients had stable angina or evidence of ischemia by stress test without elevation of cardiac enzymes. Median interval between previous intervention and presentation for clinical in-stent restenosis was shorter for patients with MI than for non-MI patients (STEMI, 90 days; NSTEMI, 79 days; non-MI, 125 days; p=0.07). Diffuse in-stent restenosis was more frequent in MI patients than in non-MI patients (72.7% vs 56.3%; p<0.005). Renal failure was more prevalent in patients with MI than in those without MI (31.8% vs 6.3%, p=0.001). Compared with the non-MI group, patients with MI were more likely to have acute coronary syndromes at the time of index procedure (81.8% vs 56.8%, p=0.02). CONCLUSION: Clinical in-stent restenosis can frequently present as MI and such patients are more likely to have an aggressive angiographic pattern of restenosis. Renal failure and acute coronary syndromes at the initial procedure are associated with MI.

Rheolytic thrombectomy for thromboembolic occlusion of the internal carotid artery complicating coronary intervention.

Stroke associated with percutaneous coronary intervention (PCI) is an infrequent, but devastating complication. Intra-arterial thrombolysis has been used with limited success, but there are no widely accepted strategies for acute ischemic stroke during PCI. We report a case in which the AngioJet rheolytic thrombectomy device was used in an attempt to re-establish patency of an occluded internal carotid artery in the setting of an acute stroke caused by thromboembolization from the right coronary artery during PCI.

Paradoxical embolism in the left main coronary artery: diagnosis by transesophageal echocardiography.

We describe a patient with a paradoxical coronary embolism diagnosed by transesophageal echocardiography. The patient developed a stroke followed by a myocardial infarction. Coronary angiography showed an obstruction of the left main coronary artery. Transesophageal echocardiography showed the mechanism of the neurologic and cardiac events to be a paradoxical embolism. Emergency surgical retrieval of the thrombus lodged in the left main coronary ostium and of a separate thrombus traversing a patent foramen ovale was performed. To our knowledge, direct visualization of the paradoxical coronary embolism by echocardiography has not been reported previously. We discuss mechanisms responsible for paradoxical coronary embolism and review the literature pertaining to this condition.